Seventeen hyperkinetic children who had previously responded to sympathomimetic amines were given three different dosages of caffeine in counterbalanced order (placebo, and low and high doses equivalent to one and three cups of coffee). One hour following ingestions they were tested, double-blind on measures of visual evoked response, alpha time, vigilance, and activity level. There was a significant effect on evoked response. The behavioral measures tended to be affected in a dose-related manner but not to a statistically significant degree. It is concluded that although centrally active, caffeine does not show the congruence between behavioral and central effects that other stimulants useful in behavioral management have shown.